Consistently great customer service. Across every conversation.
Revelir grades every conversation your team handles, against the QA scorecard you already use.
Live in production at enterprises like
Revelir grades every conversation your team handles, against the QA scorecard you already use.
Live in production at enterprises like
Score
40 / 100
Status
✦ Assessed by AIConversation
Customer · 9:42 AM
I was just charged for $55.99 for a subscription I cancelled last month! This is outrageous. I want this fixed immediately.
Michael Chen · 9:44 AM
Welcome to Revelir! I'm Michael Chen, happy to help you today. I understand you're upset about an incorrect charge of $55.99 for a subscription you believe was cancelled. I can definitely look into this for you.
Customer · 9:45 AM
Yes, that's right. I cancelled it weeks ago. You guys are thieves!
Michael Chen · 9:47 AM
I can see how frustrating that must be, Ahmed. To help me investigate this specific charge, could you please provide the transaction ID or reference number for the $55.99 payment?
Customer · 9:48 AM
Why should I have to do your job for you? Just find it! It's REF-789012.
Michael Chen · 9:50 AM
Thank you for providing the reference number, REF-789012. It looks like this charge is related to a recurring service that, according to our system, was not fully cancelled before the billing cycle. I need to raise a ticket to our L2 support team to review the cancellation request details and see if we can process a refund.
Customer · 9:52 AM
L2 support? What is that? This is unacceptable. I already told you I cancelled it. Just give me my money back!
QA scorecard · 10 metrics
Open any failed metric to see the evidence behind it.
Why this failed
correct team
"I need to raise a ticket to our L2 support team to review the cancellation request details and see if we can process a refund based on the specific circumstances."
realistic timeline
"This typically falls within our standard SLA for such issues, and they should reach out to you within 2 business days with an update or resolution."
Checked against Disputes & Chargebacks SOP v4 · Your QA lead can override this grade
Why this failed
stated a policy that does not apply
"It looks like this charge is related to a recurring service that, according to our system, was not fully cancelled before the billing cycle."
Checked against Recurring Billing SOP v2 · Your QA lead can override this grade
Why this failed
solution offered before the cause was established
"I understand you're upset about an incorrect charge of $55.99 for a subscription you believe was cancelled. I can definitely look into this for you."
Checked against Billing Dispute Handling SOP v3 · Your QA lead can override this grade
The problem
Both were trained on the same refund policy. In your dashboard both tickets look the same: resolved, on time, no complaint attached.
Says the booking is not refundable, politely, and closes the ticket. The policy said it was refundable.
Checks the policy, confirms the refund, and explains when the money will arrive.
2 to 5%
of conversations get reviewed by a human. That is the manual QA industry standard. Everything grey went out unchecked.
Adam's ticket was one of the grey squares. Nobody reviewed it, and nobody knew until the customer disputed the charge.
How it works
Our team does the setup with you.
Tickets flow into Revelir automatically as they close.
Your SOPs, refund rules, escalation paths and knowledge base, so Revelir knows what the correct answer should have been.
The QA scorecard your team already uses, metric by metric, with your own definitions.
Your QA team grades a set by hand. We compare, adjust, and keep going until Revelir matches their judgement. You see the agreement rate for yourself.
With 30 agents and tens of thousands of tickets a month, no team lead can give that feedback daily. Every agent gets their own coach, built from their own graded conversations.
We've manually reviewed tickets for years. Revelir is the first product that has made AI ticket review at scale actually usable.
The team is incredibly responsive. Feedback turns into shipped features fast, it genuinely feels like we're building the product together.
Who this is for
No. Revelir is configured against the scorecard your team already uses, metric by metric, with your own definitions. Our team does that configuration for you.
Your QA lead sees the reasoning, the quote it came from, and the document it was checked against, then overrides the grade. Those overrides are how we calibrate, so disagreement is the useful part.
Around 90% agreement with human QA assessments today. You see the agreement rate on your own conversations during the pilot, before committing to anything.
Helpdesk QA tools give your analysts a form to fill in, so coverage is still limited by how many tickets a person can read. Revelir does the grading itself, on all of them, and checks answers against your own policy documents.
We integrate through the APIs of all major helpdesks, Zendesk and Salesforce included. If your team runs on something else, tell us on the call and we will confirm it before any pilot starts.
A pilot on your own tickets runs in days. Most of that time is us reading your documents and calibrating against your graders.
Data is processed in our cloud environment and is never used to train third-party models. Full sub-processor list and security documentation available on request.
Thirty minutes to walk through how your team grades conversations today. Then we grade a set of your own, and show you where Revelir agreed with your QA team and where it did not.